antsnormflows.distributions.base

Classes

AffineGaussian(shape, affine_shape[, ...])	Diagonal Gaussian with an affine constant transform applied (optionally class conditional).
BaseDistribution()	Base distribution of a flow-based model.
ClassCondDiagGaussian(shape, num_classes[, ...])	Class-conditional diagonal Gaussian with clamped log-scales.
ConditionalDiagGaussian(shape, context_encoder)	Conditional multivariate diagonal Gaussian; mean and log sigma are produced by a context encoder.
DiagGaussian(shape[, trainable, min_log, ...])	Multivariate Gaussian with diagonal covariance.
GaussianMixture(n_modes, dim[, loc, scale, ...])	Mixture of diagonal Gaussians.
GaussianPCA(dim[, latent_dim, sigma, jitter])	Gaussian induced by a linear map of a latent (content) variable with Gaussian noise:
GlowBase(shape[, num_classes, ...])	Diagonal Gaussian with one mean and log scale per channel (Glow).
Uniform(shape[, low, high])	Multivariate uniform distribution over [low, high].
UniformGaussian(ndim, ind[, scale])	1D random variable with some entries uniform and others Gaussian.